Importing libraries¶

In [ ]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from scipy.stats import ttest_ind
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
from scipy import stats
from imblearn.over_sampling import SMOTE
from imblearn.under_sampling import RandomUnderSampler
from imblearn.pipeline import Pipeline
from sklearn.model_selection import train_test_split
from sklearn.feature_selection import SelectKBest, chi2
from sklearn.ensemble import RandomForestClassifier
import plotly as po
import plotly.express as px
import scipy.cluster.hierarchy as sch
import scipy.stats as stats
po.offline.init_notebook_mode(connected=True)

Loading the dataset¶

In [ ]:
data = pd.read_csv("./CardiacPrediction.csv")
data.head()
Out[ ]:
SEQN Gender Age Annual-Family-Income Ratio-Family-Income-Poverty X60-sec-pulse Systolic Diastolic Weight Height ... Total-Cholesterol HDL Glycohemoglobin Vigorous-work Moderate-work Health-Insurance Diabetes Blood-Rel-Diabetes Blood-Rel-Stroke CoronaryHeartDisease
0 2 1 77 8 5.00 68 98 56 75.4 174.0 ... 5.56 1.39 4.7 3 3 1 2 2 2 0
1 5 1 49 11 5.00 66 122 83 92.5 178.3 ... 7.21 1.08 5.5 1 1 1 2 2 2 0
2 12 1 37 11 4.93 64 174 99 99.2 180.0 ... 4.03 0.98 5.2 2 1 1 2 1 1 0
3 13 1 70 3 1.07 102 130 66 63.6 157.7 ... 8.12 1.28 7.6 3 3 1 1 1 2 0
4 14 1 81 5 2.67 72 136 61 75.5 166.2 ... 4.50 1.04 5.8 1 1 1 2 2 2 0

5 rows × 51 columns

EDA¶

In [ ]:
data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 37079 entries, 0 to 37078
Data columns (total 51 columns):
 #   Column                       Non-Null Count  Dtype  
---  ------                       --------------  -----  
 0   SEQN                         37079 non-null  int64  
 1   Gender                       37079 non-null  int64  
 2   Age                          37079 non-null  int64  
 3   Annual-Family-Income         37079 non-null  int64  
 4   Ratio-Family-Income-Poverty  37079 non-null  float64
 5   X60-sec-pulse                37079 non-null  int64  
 6   Systolic                     37079 non-null  int64  
 7   Diastolic                    37079 non-null  int64  
 8   Weight                       37079 non-null  float64
 9   Height                       37079 non-null  float64
 10  Body-Mass-Index              37079 non-null  float64
 11  White-Blood-Cells            37079 non-null  float64
 12  Lymphocyte                   37079 non-null  float64
 13  Monocyte                     37079 non-null  float64
 14  Eosinophils                  37079 non-null  float64
 15  Basophils                    37079 non-null  float64
 16  Red-Blood-Cells              37079 non-null  float64
 17  Hemoglobin                   37079 non-null  float64
 18  Mean-Cell-Vol                37079 non-null  float64
 19  Mean-Cell-Hgb-Conc.          37079 non-null  float64
 20  Mean-cell-Hemoglobin         37079 non-null  float64
 21  Platelet-count               37079 non-null  float64
 22  Mean-Platelet-Vol            37079 non-null  float64
 23  Segmented-Neutrophils        37079 non-null  float64
 24  Hematocrit                   37079 non-null  float64
 25  Red-Cell-Distribution-Width  37079 non-null  float64
 26  Albumin                      37079 non-null  int64  
 27  ALP                          37079 non-null  int64  
 28  AST                          37079 non-null  int64  
 29  ALT                          37079 non-null  int64  
 30  Cholesterol                  37079 non-null  float64
 31  Creatinine                   37079 non-null  float64
 32  Glucose                      37079 non-null  float64
 33  GGT                          37079 non-null  int64  
 34  Iron                         37079 non-null  float64
 35  LDH                          37079 non-null  int64  
 36  Phosphorus                   37079 non-null  float64
 37  Bilirubin                    37079 non-null  float64
 38  Protein                      37079 non-null  float64
 39  Uric.Acid                    37079 non-null  float64
 40  Triglycerides                37079 non-null  float64
 41  Total-Cholesterol            37079 non-null  float64
 42  HDL                          37079 non-null  float64
 43  Glycohemoglobin              37079 non-null  float64
 44  Vigorous-work                37079 non-null  int64  
 45  Moderate-work                37079 non-null  int64  
 46  Health-Insurance             37079 non-null  int64  
 47  Diabetes                     37079 non-null  int64  
 48  Blood-Rel-Diabetes           37079 non-null  int64  
 49  Blood-Rel-Stroke             37079 non-null  int64  
 50  CoronaryHeartDisease         37079 non-null  int64  
dtypes: float64(31), int64(20)
memory usage: 14.4 MB
In [ ]:
data.columns
Out[ ]:
Index(['SEQN', 'Gender', 'Age', 'Annual-Family-Income',
       'Ratio-Family-Income-Poverty', 'X60-sec-pulse', 'Systolic', 'Diastolic',
       'Weight', 'Height', 'Body-Mass-Index', 'White-Blood-Cells',
       'Lymphocyte', 'Monocyte', 'Eosinophils', 'Basophils', 'Red-Blood-Cells',
       'Hemoglobin', 'Mean-Cell-Vol', 'Mean-Cell-Hgb-Conc.',
       'Mean-cell-Hemoglobin', 'Platelet-count', 'Mean-Platelet-Vol',
       'Segmented-Neutrophils', 'Hematocrit', 'Red-Cell-Distribution-Width',
       'Albumin', 'ALP', 'AST', 'ALT', 'Cholesterol', 'Creatinine', 'Glucose',
       'GGT', 'Iron', 'LDH', 'Phosphorus', 'Bilirubin', 'Protein', 'Uric.Acid',
       'Triglycerides', 'Total-Cholesterol', 'HDL', 'Glycohemoglobin',
       'Vigorous-work', 'Moderate-work', 'Health-Insurance', 'Diabetes',
       'Blood-Rel-Diabetes', 'Blood-Rel-Stroke', 'CoronaryHeartDisease'],
      dtype='object')
In [ ]:
data.describe()
Out[ ]:
SEQN Gender Age Annual-Family-Income Ratio-Family-Income-Poverty X60-sec-pulse Systolic Diastolic Weight Height ... Total-Cholesterol HDL Glycohemoglobin Vigorous-work Moderate-work Health-Insurance Diabetes Blood-Rel-Diabetes Blood-Rel-Stroke CoronaryHeartDisease
count 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 ... 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000 37079.000000
mean 48901.041236 1.513282 48.943661 7.358208 2.559026 72.579250 124.090078 69.919253 80.988276 167.389601 ... 5.081713 1.370344 5.676496 1.783840 1.598856 1.218587 1.907333 1.549502 1.796165 0.040670
std 26753.636441 0.499830 18.010440 3.994083 1.624789 12.242108 19.254741 13.575804 20.678734 10.122908 ... 1.072682 0.415985 1.050223 0.448324 0.511199 0.461102 0.349674 0.497550 0.402853 0.197527
min 2.000000 1.000000 20.000000 1.000000 0.000000 32.000000 0.000000 0.000000 32.300000 129.700000 ... 1.530000 0.160000 2.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 0.000000
25% 26120.500000 1.000000 33.000000 4.000000 1.140000 64.000000 111.000000 62.000000 66.500000 160.000000 ... 4.320000 1.070000 5.200000 2.000000 1.000000 1.000000 2.000000 1.000000 2.000000 0.000000
50% 50065.000000 2.000000 48.000000 7.000000 2.180000 72.000000 121.000000 70.000000 78.200000 167.100000 ... 5.020000 1.290000 5.400000 2.000000 2.000000 1.000000 2.000000 2.000000 2.000000 0.000000
75% 71173.500000 2.000000 63.000000 10.000000 4.130000 80.000000 134.000000 78.000000 92.100000 174.600000 ... 5.740000 1.600000 5.800000 2.000000 2.000000 1.000000 2.000000 2.000000 2.000000 0.000000
max 93702.000000 2.000000 85.000000 15.000000 5.000000 224.000000 270.000000 132.000000 371.000000 204.500000 ... 14.090000 5.840000 18.800000 3.000000 3.000000 9.000000 3.000000 2.000000 2.000000 1.000000

8 rows × 51 columns

In [ ]:
#null_summary = pd.DataFrame({'Null Count': data.isnull().sum(), 'Null Percentage': (data.isnull().sum() / len(data)) * 100})

# Calculate null counts in each column
null_counts = data.isnull().sum()

# Calculate null percentage in each column
total_rows = len(data)
null_percentage = (null_counts / total_rows) * 100

# Print null counts and percentages as a table
print("Column Name                    | Null Count  | Null Percentage")
print("-" * 60)  # Separator line

for column in data.columns:
    print(f"{column:<30} | {null_counts[column]:<11} | {null_percentage[column]:.2f}%")

Column Name                    | Null Count  | Null Percentage
------------------------------------------------------------
SEQN                           | 0           | 0.00%
Gender                         | 0           | 0.00%
Age                            | 0           | 0.00%
Annual-Family-Income           | 0           | 0.00%
Ratio-Family-Income-Poverty    | 0           | 0.00%
X60-sec-pulse                  | 0           | 0.00%
Systolic                       | 0           | 0.00%
Diastolic                      | 0           | 0.00%
Weight                         | 0           | 0.00%
Height                         | 0           | 0.00%
Body-Mass-Index                | 0           | 0.00%
White-Blood-Cells              | 0           | 0.00%
Lymphocyte                     | 0           | 0.00%
Monocyte                       | 0           | 0.00%
Eosinophils                    | 0           | 0.00%
Basophils                      | 0           | 0.00%
Red-Blood-Cells                | 0           | 0.00%
Hemoglobin                     | 0           | 0.00%
Mean-Cell-Vol                  | 0           | 0.00%
Mean-Cell-Hgb-Conc.            | 0           | 0.00%
Mean-cell-Hemoglobin           | 0           | 0.00%
Platelet-count                 | 0           | 0.00%
Mean-Platelet-Vol              | 0           | 0.00%
Segmented-Neutrophils          | 0           | 0.00%
Hematocrit                     | 0           | 0.00%
Red-Cell-Distribution-Width    | 0           | 0.00%
Albumin                        | 0           | 0.00%
ALP                            | 0           | 0.00%
AST                            | 0           | 0.00%
ALT                            | 0           | 0.00%
Cholesterol                    | 0           | 0.00%
Creatinine                     | 0           | 0.00%
Glucose                        | 0           | 0.00%
GGT                            | 0           | 0.00%
Iron                           | 0           | 0.00%
LDH                            | 0           | 0.00%
Phosphorus                     | 0           | 0.00%
Bilirubin                      | 0           | 0.00%
Protein                        | 0           | 0.00%
Uric.Acid                      | 0           | 0.00%
Triglycerides                  | 0           | 0.00%
Total-Cholesterol              | 0           | 0.00%
HDL                            | 0           | 0.00%
Glycohemoglobin                | 0           | 0.00%
Vigorous-work                  | 0           | 0.00%
Moderate-work                  | 0           | 0.00%
Health-Insurance               | 0           | 0.00%
Diabetes                       | 0           | 0.00%
Blood-Rel-Diabetes             | 0           | 0.00%
Blood-Rel-Stroke               | 0           | 0.00%
CoronaryHeartDisease           | 0           | 0.00%
In [ ]:
# Calculate the number of distinct values in each column
distinct_counts = data.nunique()

# Print the number of distinct values and distinct values in each column
print("Column Name                    | Distinct Count ")
print("-" * 50)  # Separator line

for column in data.columns:
    distinct_values = data[column].unique()
    print(f"{column:<30} | {distinct_counts[column]:<14}")

Column Name                    | Distinct Count 
--------------------------------------------------
SEQN                           | 37079         
Gender                         | 2             
Age                            | 66            
Annual-Family-Income           | 13            
Ratio-Family-Income-Poverty    | 501           
X60-sec-pulse                  | 60            
Systolic                       | 154           
Diastolic                      | 114           
Weight                         | 1432          
Height                         | 627           
Body-Mass-Index                | 3235          
White-Blood-Cells              | 212           
Lymphocyte                     | 610           
Monocyte                       | 218           
Eosinophils                    | 218           
Basophils                      | 70            
Red-Blood-Cells                | 404           
Hemoglobin                     | 131           
Mean-Cell-Vol                  | 538           
Mean-Cell-Hgb-Conc.            | 247           
Mean-cell-Hemoglobin           | 4501          
Platelet-count                 | 551           
Mean-Platelet-Vol              | 82            
Segmented-Neutrophils          | 677           
Hematocrit                     | 329           
Red-Cell-Distribution-Width    | 156           
Albumin                        | 39            
ALP                            | 262           
AST                            | 202           
ALT                            | 224           
Cholesterol                    | 543           
Creatinine                     | 371           
Glucose                        | 616           
GGT                            | 387           
Iron                           | 493           
LDH                            | 312           
Phosphorus                     | 101           
Bilirubin                      | 77            
Protein                        | 57            
Uric.Acid                      | 122           
Triglycerides                  | 798           
Total-Cholesterol              | 313           
HDL                            | 276           
Glycohemoglobin                | 127           
Vigorous-work                  | 3             
Moderate-work                  | 3             
Health-Insurance               | 4             
Diabetes                       | 3             
Blood-Rel-Diabetes             | 2             
Blood-Rel-Stroke               | 2             
CoronaryHeartDisease           | 2
In [ ]:
data[data['Gender']==1].count()
Out[ ]:
SEQN                           18047
Gender                         18047
Age                            18047
Annual-Family-Income           18047
Ratio-Family-Income-Poverty    18047
X60-sec-pulse                  18047
Systolic                       18047
Diastolic                      18047
Weight                         18047
Height                         18047
Body-Mass-Index                18047
White-Blood-Cells              18047
Lymphocyte                     18047
Monocyte                       18047
Eosinophils                    18047
Basophils                      18047
Red-Blood-Cells                18047
Hemoglobin                     18047
Mean-Cell-Vol                  18047
Mean-Cell-Hgb-Conc.            18047
Mean-cell-Hemoglobin           18047
Platelet-count                 18047
Mean-Platelet-Vol              18047
Segmented-Neutrophils          18047
Hematocrit                     18047
Red-Cell-Distribution-Width    18047
Albumin                        18047
ALP                            18047
AST                            18047
ALT                            18047
Cholesterol                    18047
Creatinine                     18047
Glucose                        18047
GGT                            18047
Iron                           18047
LDH                            18047
Phosphorus                     18047
Bilirubin                      18047
Protein                        18047
Uric.Acid                      18047
Triglycerides                  18047
Total-Cholesterol              18047
HDL                            18047
Glycohemoglobin                18047
Vigorous-work                  18047
Moderate-work                  18047
Health-Insurance               18047
Diabetes                       18047
Blood-Rel-Diabetes             18047
Blood-Rel-Stroke               18047
CoronaryHeartDisease           18047
dtype: int64
In [ ]:
data[data['Gender']==2].count()
Out[ ]:
SEQN                           19032
Gender                         19032
Age                            19032
Annual-Family-Income           19032
Ratio-Family-Income-Poverty    19032
X60-sec-pulse                  19032
Systolic                       19032
Diastolic                      19032
Weight                         19032
Height                         19032
Body-Mass-Index                19032
White-Blood-Cells              19032
Lymphocyte                     19032
Monocyte                       19032
Eosinophils                    19032
Basophils                      19032
Red-Blood-Cells                19032
Hemoglobin                     19032
Mean-Cell-Vol                  19032
Mean-Cell-Hgb-Conc.            19032
Mean-cell-Hemoglobin           19032
Platelet-count                 19032
Mean-Platelet-Vol              19032
Segmented-Neutrophils          19032
Hematocrit                     19032
Red-Cell-Distribution-Width    19032
Albumin                        19032
ALP                            19032
AST                            19032
ALT                            19032
Cholesterol                    19032
Creatinine                     19032
Glucose                        19032
GGT                            19032
Iron                           19032
LDH                            19032
Phosphorus                     19032
Bilirubin                      19032
Protein                        19032
Uric.Acid                      19032
Triglycerides                  19032
Total-Cholesterol              19032
HDL                            19032
Glycohemoglobin                19032
Vigorous-work                  19032
Moderate-work                  19032
Health-Insurance               19032
Diabetes                       19032
Blood-Rel-Diabetes             19032
Blood-Rel-Stroke               19032
CoronaryHeartDisease           19032
dtype: int64
In [ ]:
In [ ]:
data['Gender'].value_counts()
Out[ ]:
Gender
2    19032
1    18047
Name: count, dtype: int64
In [ ]:
data['CoronaryHeartDisease'].value_counts()
Out[ ]:
CoronaryHeartDisease
0    35571
1     1508
Name: count, dtype: int64

Balancinng the data using SMOTE¶

In [ ]:
# Assuming 'data' is your DataFrame
# X should contain your feature columns, and y should contain your target variable

# Separate features (X) and target variable (y)
X = data.drop('CoronaryHeartDisease', axis=1)
y = data['CoronaryHeartDisease']

# Define the pipeline with SMOTE and RandomUnderSampler
pipeline = Pipeline([
    ('under_sampler', RandomUnderSampler(sampling_strategy=0.10, random_state=42)),  # 4:1 undersampling
    ('smote', SMOTE(sampling_strategy='auto', random_state=42))  # Auto-adjusted oversampling
])



# Fit and transform the data using the pipeline
X_resampled, y_resampled = pipeline.fit_resample(X, y)

# Create a new DataFrame with the resampled data
resampled_data = pd.concat([pd.DataFrame(X_resampled, columns=X.columns), pd.Series(y_resampled, name='CoronaryHeartDisease')], axis=1)
In [ ]:
resampled_data.head()
Out[ ]:
SEQN Gender Age Annual-Family-Income Ratio-Family-Income-Poverty X60-sec-pulse Systolic Diastolic Weight Height ... Total-Cholesterol HDL Glycohemoglobin Vigorous-work Moderate-work Health-Insurance Diabetes Blood-Rel-Diabetes Blood-Rel-Stroke CoronaryHeartDisease
0 61989 1 46 10 2.71 64 114 74 102.1 185.5 ... 5.84 1.71 5.3 1 1 1 2 2 2 0
1 47788 2 75 7 2.56 58 120 70 69.5 165.6 ... 3.62 1.32 5.1 2 2 1 2 1 2 0
2 68910 1 74 7 1.70 76 138 80 86.2 170.4 ... 4.03 0.83 5.9 2 2 1 2 2 2 0
3 24195 2 46 11 4.18 74 122 85 90.1 178.9 ... 6.75 1.34 5.1 2 2 1 2 2 1 0
4 52532 1 80 7 2.40 82 92 56 88.3 181.9 ... 3.39 1.71 6.4 2 2 1 1 1 2 0

5 rows × 51 columns

In [ ]:
resampled_data.shape
Out[ ]:
(30160, 51)
In [ ]:
resampled_data['CoronaryHeartDisease'].value_counts()
Out[ ]:
CoronaryHeartDisease
0    15080
1    15080
Name: count, dtype: int64
In [ ]:
resampled_data.columns
Out[ ]:
Index(['SEQN', 'Gender', 'Age', 'Annual-Family-Income',
       'Ratio-Family-Income-Poverty', 'X60-sec-pulse', 'Systolic', 'Diastolic',
       'Weight', 'Height', 'Body-Mass-Index', 'White-Blood-Cells',
       'Lymphocyte', 'Monocyte', 'Eosinophils', 'Basophils', 'Red-Blood-Cells',
       'Hemoglobin', 'Mean-Cell-Vol', 'Mean-Cell-Hgb-Conc.',
       'Mean-cell-Hemoglobin', 'Platelet-count', 'Mean-Platelet-Vol',
       'Segmented-Neutrophils', 'Hematocrit', 'Red-Cell-Distribution-Width',
       'Albumin', 'ALP', 'AST', 'ALT', 'Cholesterol', 'Creatinine', 'Glucose',
       'GGT', 'Iron', 'LDH', 'Phosphorus', 'Bilirubin', 'Protein', 'Uric.Acid',
       'Triglycerides', 'Total-Cholesterol', 'HDL', 'Glycohemoglobin',
       'Vigorous-work', 'Moderate-work', 'Health-Insurance', 'Diabetes',
       'Blood-Rel-Diabetes', 'Blood-Rel-Stroke', 'CoronaryHeartDisease'],
      dtype='object')

Feature Selection¶

In [ ]:
# Assuming your dataset has features (X) and labels (y)
# Separate the features and the target variable
X = resampled_data.drop(columns=['CoronaryHeartDisease', 'SEQN','Health-Insurance','Cholesterol'])  # Features
y = resampled_data['CoronaryHeartDisease']  # Target variable

# Split the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Step 1: Apply SelectKBest
k_best_selector = SelectKBest(chi2, k=30)  # Adjust k as needed
X_train_k_best = k_best_selector.fit_transform(X_train, y_train)
selected_features = X.columns[k_best_selector.get_support()]

# Step 2: Apply RandomForestClassifier for further feature importance
rf_classifier = RandomForestClassifier(n_estimators=100, random_state=42)
rf_classifier.fit(X_train_k_best, y_train)

# Display feature importance scores
feature_importances = pd.DataFrame(rf_classifier.feature_importances_,
                                   index=selected_features,
                                   columns=['importance']).sort_values('importance', ascending=False)

print("Selected Features:")
print(selected_features)
print("\nFeature Importance Scores:")
print(feature_importances)

Selected Features:
Index(['Gender', 'Age', 'Annual-Family-Income', 'X60-sec-pulse', 'Systolic',
       'Diastolic', 'Weight', 'Lymphocyte', 'Monocyte', 'Eosinophils',
       'Mean-Cell-Vol', 'Platelet-count', 'Segmented-Neutrophils', 'Albumin',
       'ALP', 'ALT', 'Creatinine', 'Glucose', 'GGT', 'Iron', 'LDH',
       'Bilirubin', 'Uric.Acid', 'Triglycerides', 'Total-Cholesterol',
       'Glycohemoglobin', 'Moderate-work', 'Diabetes', 'Blood-Rel-Diabetes',
       'Blood-Rel-Stroke'],
      dtype='object')

Feature Importance Scores:
                       importance
Age                      0.205972
Gender                   0.094800
Diabetes                 0.066087
Creatinine               0.048159
Glycohemoglobin          0.045138
Glucose                  0.033609
Total-Cholesterol        0.030954
Blood-Rel-Stroke         0.029520
Diastolic                0.027677
Triglycerides            0.026348
Albumin                  0.025268
X60-sec-pulse            0.023563
Uric.Acid                0.023294
Platelet-count           0.022489
LDH                      0.021782
Bilirubin                0.021246
Lymphocyte               0.021055
Blood-Rel-Diabetes       0.020636
Weight                   0.020018
Systolic                 0.019860
Eosinophils              0.019822
Mean-Cell-Vol            0.019739
Monocyte                 0.019480
Annual-Family-Income     0.018784
Iron                     0.018568
Segmented-Neutrophils    0.017016
ALT                      0.016564
ALP                      0.016124
GGT                      0.016045
Moderate-work            0.010383
In [ ]:
# Visualize feature importance with a red line indicating the threshold
threshold = 0.0225 # Adjust the threshold as needed
plt.figure(figsize=(20, 10))
ax = plt.barh(feature_importances.index, feature_importances['importance'], color='skyblue')
plt.axvline(x=threshold, color='red', linestyle='--', label=f'Threshold ({threshold})')
plt.xlabel('Importance Score')
plt.title('Feature Importance Scores')
plt.legend()
plt.show()
No description has been provided for this image
In [ ]:
# Print features above the threshold line
selected_features_above_threshold = feature_importances[feature_importances['importance'] > threshold].index
print("\nFeatures Above the Threshold:")
print(selected_features_above_threshold)
print(len(selected_features_above_threshold))

Features Above the Threshold:
Index(['Age', 'Gender', 'Diabetes', 'Creatinine', 'Glycohemoglobin', 'Glucose',
       'Total-Cholesterol', 'Blood-Rel-Stroke', 'Diastolic', 'Triglycerides',
       'Albumin', 'X60-sec-pulse', 'Uric.Acid'],
      dtype='object')
13
In [ ]:
features=list(selected_features_above_threshold)
print(features)
features.append('CoronaryHeartDisease')
final_data=data[features]
final_data.head()

['Age', 'Gender', 'Diabetes', 'Creatinine', 'Glycohemoglobin', 'Glucose', 'Total-Cholesterol', 'Blood-Rel-Stroke', 'Diastolic', 'Triglycerides', 'Albumin', 'X60-sec-pulse', 'Uric.Acid']
Out[ ]:
Age Gender Diabetes Creatinine Glycohemoglobin Glucose Total-Cholesterol Blood-Rel-Stroke Diastolic Triglycerides Albumin X60-sec-pulse Uric.Acid CoronaryHeartDisease
0 77 1 2 61.9 4.7 4.330 5.56 2 56 1.298 45 68 362.8 0
1 49 1 2 70.7 5.5 5.273 7.21 2 83 3.850 45 66 404.5 0
2 37 1 2 88.4 5.2 4.163 4.03 1 99 1.581 47 64 339.0 0
3 70 1 1 61.9 7.6 7.882 8.12 2 66 3.635 40 102 410.4 0
4 81 1 2 88.4 5.8 6.384 4.50 2 61 0.756 45 72 368.8 0

Data Analysis¶

In [ ]:
sns.histplot(data=data, x='Age',hue='CoronaryHeartDisease',kde=True)
Out[ ]:
<Axes: xlabel='Age', ylabel='Count'>
No description has been provided for this image
In [ ]:
# Filter the DataFrame for each gender
male_data = data[data['Gender'] == 1]
female_data = data[data['Gender'] == 2]

# Create separate figures for each gender's pie chart
fig, axes = plt.subplots(1, 2, figsize=(12, 6))

# Plot pie chart for Male data
male_counts = male_data['CoronaryHeartDisease'].value_counts()
axes[0].pie(male_counts, labels=male_counts.index, autopct='%1.1f%%', startangle=90)
axes[0].set_title('Male - CoronaryHeartDisease Distribution')

# Plot pie chart for Female data
female_counts = female_data['CoronaryHeartDisease'].value_counts()
axes[1].pie(female_counts, labels=female_counts.index, autopct='%1.1f%%', startangle=90)
axes[1].set_title('Female - CoronaryHeartDisease Distribution')

plt.tight_layout()
plt.show()
No description has been provided for this image
In [ ]:
# Selecting the features for the correlation matrix
features = ['Age', 'Gender', 'Diabetes', 'Creatinine', 'Glycohemoglobin', 'Glucose',
            'Total-Cholesterol', 'Blood-Rel-Stroke', 'Diastolic', 'Triglycerides',
            'Albumin', 'X60-sec-pulse', 'Uric.Acid', 'CoronaryHeartDisease']

# Create a correlation matrix
corr_matrix = final_data.corr()

# Set up the matplotlib figure
fig, ax = plt.subplots(figsize=(12, 10))

# Create a heatmap with annotations
sns.heatmap(corr_matrix, annot=True, cmap='coolwarm', fmt=".2f", ax=ax)

# Show the plot
plt.show()
No description has been provided for this image
In [ ]:
# Selecting the features for which you want to plot the dendrogram
features = ['Age', 'Gender', 'Diabetes', 'Creatinine', 'Glycohemoglobin', 'Glucose',
            'Total-Cholesterol', 'Blood-Rel-Stroke', 'Diastolic', 'Triglycerides',
            'Albumin', 'X60-sec-pulse', 'Uric.Acid']

# Create a correlation matrix
corr_matrix = final_data.corr()

# Create a dendrogram
dendrogram = sch.dendrogram(sch.linkage(corr_matrix, method='ward'), labels=corr_matrix.columns, orientation='right')

# Add title and show the plot
plt.title('Dendrogram')
plt.show()
No description has been provided for this image
In [ ]:
# Normalize the 'Triglycerides' values for bubble sizes
normalized_triglycerides = (final_data['Triglycerides'] - final_data['Triglycerides'].min()) / (final_data['Triglycerides'].max() - final_data['Triglycerides'].min())

# Create an interactive scatter plot using Plotly Express
fig = px.scatter(final_data, x='Age', y='Uric.Acid', size=normalized_triglycerides, color='CoronaryHeartDisease')

# Update layout for better interactivity
fig.update_layout(
    title='Interactive Scatter Plot',
    xaxis_title='Age',
    yaxis_title='Uric Acid',
    legend_title='Coronary Heart Disease',
    width=900,  # Set the width of the plot
    height=600,  # Set the height of the plot
    template='plotly_dark',  # Set the template to 'ggplot2'
)

# Show the interactive plot
fig.show()
In [ ]:
# Create an interactive strip plot with Gender
fig = px.strip(final_data, color='CoronaryHeartDisease', x='Creatinine', y='Uric.Acid',
               #category_orders={'Gender': [1, 2]},  # Specify the order of categories
               title='Interactive Categorical Scatter Plot',
               labels={'Total-Cholesterol': 'Total-Cholesterol', 'CoronaryHeartDisease': 'Heart Disease'},
               width=800, height=500)

# Show the plot
fig.show()
In [ ]:
# Line Plot
sns.lineplot(x='Age', y='Total-Cholesterol', hue='CoronaryHeartDisease', data=final_data)
plt.show()
No description has been provided for this image

Hypothesis Testing¶

Hypothesis 1 (T-Test):

Null Hypothesis (H0): The mean age of individuals with CoronaryHeartDisease is equal to the mean age of individuals without CoronaryHeartDisease.

Alternative Hypothesis (H1): The mean age of individuals with CoronaryHeartDisease is different from the mean age of individuals without CoronaryHeartDisease.

Test: Independent samples t-test comparing the age distribution between the two groups. Hypothesis 2 (Prop Test):

In [ ]:
# Hypothesis 1: T-Test for Age
alpha = 0.05
age_with_disease = data.loc[data['CoronaryHeartDisease'] == 1, 'Age']
age_without_disease = data.loc[data['CoronaryHeartDisease'] == 0, 'Age']

t_stat_age, p_value_age = stats.ttest_ind(age_with_disease, age_without_disease)

if p_value_age < alpha:
    print("Hypothesis 1 Result: Reject the null hypothesis. The mean age is significantly different between individuals with and without CoronaryHeartDisease.")
else:
    print("Hypothesis 1 Result: Fail to reject the null hypothesis. There is no significant difference in mean age between individuals with and without CoronaryHeartDisease.")

Hypothesis 1 Result: Reject the null hypothesis. The mean age is significantly different between individuals with and without CoronaryHeartDisease.
In [ ]:
mean_age_by_chd = final_data.groupby('CoronaryHeartDisease')['Age'].mean()

# Plotting the bar plot
plt.figure(figsize=(8, 6))
mean_age_by_chd.plot(kind='bar', color='skyblue')
plt.title('Mean Age of Individuals with and without Coronary Heart Disease')
plt.xlabel('Coronary Heart Disease')
plt.ylabel('Mean Age')
plt.xticks(ticks=[0, 1], labels=['Without CHD', 'With CHD'])
plt.grid(axis='y')
plt.show()
No description has been provided for this image

Null Hypothesis (H0): The mean Glycohemoglobin level is equal in individuals with and without CoronaryHeartDisease.

Alternative Hypothesis (H1): The mean Glycohemoglobin level is different in individuals with and without CoronaryHeartDisease.

Test: Independent samples t-test comparing Glycohemoglobin levels between the two groups.

In [ ]:
# Hypothesis 2: T-Test for Glycohemoglobin
glycohemoglobin_with_disease = data.loc[data['CoronaryHeartDisease'] == 1, 'Glycohemoglobin']
glycohemoglobin_without_disease = data.loc[data['CoronaryHeartDisease'] == 0, 'Glycohemoglobin']

t_stat_glycohemoglobin, p_value_glycohemoglobin = stats.ttest_ind(glycohemoglobin_with_disease, glycohemoglobin_without_disease)

if p_value_glycohemoglobin < alpha:
    print("Hypothesis 4 Result: Reject the null hypothesis. The mean Glycohemoglobin level is significantly different between individuals with and without CoronaryHeartDisease.")
else:
    print("Hypothesis 4 Result: Fail to reject the null hypothesis. There is no significant difference in mean Glycohemoglobin level between individuals with and without CoronaryHeartDisease.")

Hypothesis 4 Result: Reject the null hypothesis. The mean Glycohemoglobin level is significantly different between individuals with and without CoronaryHeartDisease.
In [ ]:
# Interactive Scatter Plot with Trend Line
fig = px.scatter(final_data, x='Glycohemoglobin', y='Glucose', color='CoronaryHeartDisease', trendline='ols')
fig.show()

Hypothesis test for Pulse rate and Blood rel stroke

H0: Having a history of Coronary Heart disease does not have any affect on the pulse rate of people.

H_A: Having a history of Coronary Heart disease has an affect on the pulse rate of people.

In [ ]:
#Hypothesis 3: CHD and X60 pulse rate

import pandas as pd
from scipy.stats import mannwhitneyu

# Assuming 'df' is your DataFrame containing 'X60-sec-pulse' and 'Blood-Rel-Stroke' columns
# Replace 'X60-sec-pulse' and 'Blood-Rel-Stroke' with the actual column names in your DataFrame

# Separate pulse rates for individuals with and without a family history of stroke
pulse_with_stroke = final_data[final_data['Blood-Rel-Stroke'] == 1]['X60-sec-pulse']
pulse_without_stroke = final_data[final_data['Blood-Rel-Stroke'] == 2]['X60-sec-pulse']

# Perform Mann-Whitney U test
statistic, p_value = mannwhitneyu(pulse_with_stroke, pulse_without_stroke)

# Display test results
print(f"Mann-Whitney U Statistic: {statistic}")
print(f"P-value: {p_value}")

# Compare the p-value with the significance level (e.g., 0.05) to draw conclusions
if p_value<0.05:
  print("The Null Hypothesis is rejected, Having a history of Coronary Heart disease has an affect on the pulse rate of people.")
else:
  print("The Null Hypothesis is accepted, Having a history of Coronary Heart disease does not have an affect on the pulse rate of people.")

Mann-Whitney U Statistic: 115365137.5
P-value: 4.452551164503746e-06
The Null Hypothesis is rejected, Having a history of Coronary Heart disease has an affect on the pulse rate of people.
In [ ]:
sns.lmplot(data=final_data,x='Age',y='X60-sec-pulse',hue='CoronaryHeartDisease', col='Blood-Rel-Stroke',height=5)
Out[ ]:
<seaborn.axisgrid.FacetGrid at 0x2452cb125d0>
No description has been provided for this image

Hypothesis about diabeties:

H0: The proportion of people suffering with CoronaryHeartDisease is same across all the 3 diabeties types.

H_A: The proportion of people suffering with CoronaryHeartDisease is higher with Diabeties type 2 patients than the rest.

In [ ]:
# Hypothesis 4: Chi2 test for Diabeties

import pandas as pd
from scipy.stats import chi2_contingency

# Create a contingency table
contingency_table = pd.crosstab(final_data['Diabetes'], final_data['CoronaryHeartDisease'])

# Perform the Chi-square test
chi2, p_value, _, _ = chi2_contingency(contingency_table)

# Significance level (alpha)
alpha = 0.05

# Print the results
print(f'Chi-square Test for Independence between Diabetes Type and CoronaryHeartDisease:')
print(f'p-value = {p_value}')

# Check if the p-value is less than alpha to make a decision
if p_value < alpha:
    print('Reject the null hypothesis. There is evidence to suggest that the proportion of people suffering from CoronaryHeartDisease is different across diabetes types.')
else:
    print('Fail to reject the null hypothesis. There is not enough evidence to suggest a difference in the proportion of people suffering from CoronaryHeartDisease across diabetes types.')

Chi-square Test for Independence between Diabetes Type and CoronaryHeartDisease:
p-value = 1.6655000663136123e-148
Reject the null hypothesis. There is evidence to suggest that the proportion of people suffering from CoronaryHeartDisease is different across diabetes types.
In [ ]:
# Calculate proportions of CHD occurrences for each type of Diabetes
proportions = final_data.groupby('Diabetes')['CoronaryHeartDisease'].mean()

# Plotting the proportions
plt.figure(figsize=(8, 6))
proportions.plot(kind='bar', color='skyblue')
plt.title('Proportion of CHD across Diabetes Types')
plt.xlabel('Diabetes Types')
plt.ylabel('Proportion of CHD')
plt.xticks(rotation=0)
plt.grid(axis='y')
plt.axhline(proportions[2], color='red', linestyle='--', label='Diabetes Type 2 Proportion')
plt.legend()
plt.show()
No description has been provided for this image
In [ ]:
# Violin Plot of Glucose by Diabetes Status
plt.figure(figsize=(12, 8))
sns.violinplot(x='Diabetes',hue='CoronaryHeartDisease', y='Glucose', data=final_data,legend=True,split=True)
plt.title('Violin Plot of Glucose by Diabetes Status')
plt.show()
No description has been provided for this image 
# This is formatted as code

Hypothesis about X60 pulse rate and age:

H0: The pulse rate varies the same for people with and without Coronary Heart Disease.

H_A: The pulse rate varies the rigorously for people with Coronary Heart Disease than people without any heart disease.

In [ ]:
#Hypothesis 5:

from scipy.stats import ttest_ind

# Example data (replace with your actual dataset)
group_with_CHD = data[data['CoronaryHeartDisease'] == 1]['X60-sec-pulse']
group_without_CHD = data[data['CoronaryHeartDisease'] == 0]['X60-sec-pulse']

# Perform the t-test
stat, p_value = ttest_ind(group_with_CHD, group_without_CHD, equal_var=False)

# Significance level (alpha)
alpha = 0.05

# Print the results
print(f't-Test for X60 pulse rate and CoronaryHeartDisease:')
print(f'p-value = {p_value}')

# Check if the p-value is less than alpha to make a decision
if p_value < alpha:
    print('Reject the null hypothesis. There is evidence to suggest that the pulse rate varies rigorously for people with CoronaryHeartDisease than people without any heart disease.')
else:
    print('Fail to reject the null hypothesis. There is not enough evidence to suggest a difference in the variation of pulse rate between people with and without CoronaryHeartDisease.')

t-Test for X60 pulse rate and CoronaryHeartDisease:
p-value = 2.155744320482476e-41
Reject the null hypothesis. There is evidence to suggest that the pulse rate varies rigorously for people with CoronaryHeartDisease than people without any heart disease.
In [ ]:
# Line Plot
sns.lineplot(x='Age', y='X60-sec-pulse', hue='CoronaryHeartDisease', data=final_data)
plt.show()
No description has been provided for this image
In [ ]:
resampled_data.columns
Out[ ]:
Index(['SEQN', 'Gender', 'Age', 'Annual-Family-Income',
       'Ratio-Family-Income-Poverty', 'X60-sec-pulse', 'Systolic', 'Diastolic',
       'Weight', 'Height', 'Body-Mass-Index', 'White-Blood-Cells',
       'Lymphocyte', 'Monocyte', 'Eosinophils', 'Basophils', 'Red-Blood-Cells',
       'Hemoglobin', 'Mean-Cell-Vol', 'Mean-Cell-Hgb-Conc.',
       'Mean-cell-Hemoglobin', 'Platelet-count', 'Mean-Platelet-Vol',
       'Segmented-Neutrophils', 'Hematocrit', 'Red-Cell-Distribution-Width',
       'Albumin', 'ALP', 'AST', 'ALT', 'Cholesterol', 'Creatinine', 'Glucose',
       'GGT', 'Iron', 'LDH', 'Phosphorus', 'Bilirubin', 'Protein', 'Uric.Acid',
       'Triglycerides', 'Total-Cholesterol', 'HDL', 'Glycohemoglobin',
       'Vigorous-work', 'Moderate-work', 'Health-Insurance', 'Diabetes',
       'Blood-Rel-Diabetes', 'Blood-Rel-Stroke', 'CoronaryHeartDisease'],
      dtype='object')
In [ ]:
selected_features_above_threshold
Out[ ]:
Index(['Age', 'Gender', 'Diabetes', 'Creatinine', 'Glycohemoglobin', 'Glucose',
       'Total-Cholesterol', 'Blood-Rel-Stroke', 'Diastolic', 'Triglycerides',
       'Albumin', 'X60-sec-pulse', 'Uric.Acid'],
      dtype='object')
In [ ]:
features=list(selected_features_above_threshold)
print(features)

['Age', 'Gender', 'Diabetes', 'Creatinine', 'Glycohemoglobin', 'Glucose', 'Total-Cholesterol', 'Blood-Rel-Stroke', 'Diastolic', 'Triglycerides', 'Albumin', 'X60-sec-pulse', 'Uric.Acid']
In [ ]:
features.append('CoronaryHeartDisease')
final_data=resampled_data[features]
final_data.head()
Out[ ]:
Age Gender Diabetes Creatinine Glycohemoglobin Glucose Total-Cholesterol Blood-Rel-Stroke Diastolic Triglycerides Albumin X60-sec-pulse Uric.Acid CoronaryHeartDisease
0 46 1 2 94.59 5.3 4.66 5.84 2 74 0.858 42 64 237.9 0
1 75 2 2 81.33 5.1 5.00 3.62 2 70 1.840 35 58 243.9 0
2 74 1 2 60.11 5.9 5.22 4.03 2 80 3.252 41 76 285.5 0
3 46 2 2 70.72 5.1 4.88 6.75 1 85 1.118 44 74 362.8 0
4 80 1 1 137.02 6.4 6.72 3.39 2 56 0.610 41 82 440.2 0

Model Training and Evaluation¶

In [ ]:
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score, roc_auc_score, roc_curve, f1_score
import matplotlib.pyplot as plt
In [ ]:
# Assume X contains features and y contains the target variable
X = resampled_data.drop('CoronaryHeartDisease', axis=1)
y = resampled_data['CoronaryHeartDisease']

# Split the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Standardize features
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

# Initialize models
svm_model = SVC(kernel='linear', random_state=42)
rf_model = RandomForestClassifier(n_estimators=100, random_state=42)
lr_model = LogisticRegression(random_state=42)
gb_model = GradientBoostingClassifier(n_estimators=100, random_state=42)
knn_model = KNeighborsClassifier(n_neighbors=5)

# Train and predict using each model
models = [svm_model, rf_model, lr_model, gb_model, knn_model]
model_names = ['SVM', 'Random Forest', 'Logistic Regression', 'Gradient Boosting', 'KNN']
accuracies = []
roc_auc_scores = []
f1_scores = []
auc_scores = []

plt.figure(figsize=(8, 6))

for model in models:
    if model == knn_model:
        model.fit(X_train_scaled, y_train)
        predictions = model.predict(X_test_scaled)
    else:
        model.fit(X_train, y_train)
        predictions = model.predict(X_test)

    accuracy = accuracy_score(y_test, predictions)
    accuracies.append(accuracy)

    f1 = f1_score(y_test, predictions)
    f1_scores.append(f1)

    auc = roc_auc_score(y_test, predictions)
    auc_scores.append(auc)

    fpr, tpr, _ = roc_curve(y_test, predictions)
    plt.plot(fpr, tpr, label=model_names[models.index(model)])

# Labels and Legend
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('ROC Curves for Different Models')
plt.legend()
plt.grid(True)
plt.show()

/usr/local/lib/python3.10/dist-packages/sklearn/linear_model/_logistic.py:458: ConvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
    https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression
  n_iter_i = _check_optimize_result(
No description has been provided for this image
In [ ]:
# Line chart for accuracies
plt.figure(figsize=(10, 6))
plt.plot(model_names, accuracies, marker='o', linestyle='-', color='b')
plt.title('Model Accuracies')
plt.xlabel('Models')
plt.ylabel('Accuracy')
plt.ylim(0, 1.0)
plt.grid(True)
plt.show()

for i, model_name in enumerate(model_names):
    print(f'{model_name} - Accuracy: {accuracies[i]}')
No description has been provided for this image 
SVM - Accuracy: 0.8474801061007957
Random Forest - Accuracy: 0.9121352785145889
Logistic Regression - Accuracy: 0.8234416445623343
Gradient Boosting - Accuracy: 0.875
KNN - Accuracy: 0.8851127320954907
In [ ]:
# Line chart for F1 scores
plt.figure(figsize=(10, 6))
plt.plot(model_names, f1_scores, marker='o', linestyle='-', color='b')
plt.title('Model F1 Scores')
plt.xlabel('Models')
plt.ylabel('F1 Score')
plt.ylim(0, 1.0)
plt.grid(True)
plt.show()

for i, model_name in enumerate(model_names):
    print(f'{model_name} - F1 score: {f1_scores[i]}')
No description has been provided for this image 
SVM - F1 score: 0.8494764397905759
Random Forest - F1 score: 0.9133420536298233
Logistic Regression - F1 score: 0.8257241040746195
Gradient Boosting - F1 score: 0.8766361256544503
KNN - F1 score: 0.8911917098445596